Look around you. You’re probably surrounded by some kind of infrastructure, whether it’s the walls of your home or your office, the rails under the train you’re commuting on, or the sidewalk beneath your feet–by the way, please look where you’re going.
All this infrastructure has to be managed by someone. You already know how much work it takes to keep a house or apartment in good condition. Imagine how complicated it gets when you ramp that effort up to enterprise scale. Companies that work in asset-intensive industries–utilities, railways, construction companies and manufacturers for example–should inspect their infrastructure continuously to monitor the condition of their assets and conduct repairs or replacements when necessary.
These organizations devote huge amounts of resources to managing physical assets. Many have teams with acres of domain expertise built up over years of practice. At the same time, as artificial intelligence (AI) becomes more commonplace, these companies want to explore how they can use new techniques to make asset management more efficient.
At BrainCreators we bridge the gap between planning and reality allowing domain experts to apply intelligent automation for asset inspection. And we use IBM Power Systems to do it; let us tell you how.
Choosing the right starting point
BrainMatter uses the latest developments in AI to apply end-to-end intelligent automation for effective physical asset monitoring and control. Many businesses are exploring AI by experimenting with data science tooling. As a result, lots of AI models are developed that never make it to production. BrainMatter goes beyond AI experiments to provide a platform for intelligent automation of tasks that follow a perceive-decide-respond pattern.
We combine rapid recognition, direct assessment and immediate action to enable an intuitive and transparent workflow. A BrainMatter user can automate thousands of tasks without ever needing to consult a data scientist.
Behind the scenes
To build BrainMatter, we scoured the market for the best accelerated servers and tested them. IBM Power Systems was the clear leader, combining NVIDIA NVLink with IBM CPU technology to outperform competitors. Using the same technology that powers the world’s fastest supercomputers Sierra and Summit, we knew that Power Systems was the right platform to make short work of highly data-intensive use cases.
In designing our architecture, we replicated the IBM-recommended approach, opting for a containerized design built on Linux and orchestrated by Kubernetes.
With BrainMatter up and running, we can now construct AI models for asset inspection, monitoring and control, quickly and easily. With the powerful IBM server platform, we can rapidly train the models with millions of images, enabling them to identify patterns and make accurate predictions about asset condition.
BrainMatter in action
Our platform is exceptionally good at turning domain expertise and skills into intelligent automation to make life easier for your employees and optimize your processes. It’s designed to perceive through rapid recognition, decide by instant reality check against norm or plan, and respond with immediate action.
For example, we just announced a partnership with Unihorn to intelligently automate their road inspection solution. Until now, they have relied on cars equipped with cameras to drive thousands of kilometers of road. A team of road inspectors analyzes the footage over the course of five weeks, classifying defects that deviate from the norm, raising tickets and determining the next best action. With BrainMatter, they’ll shrink the road inspection process from five weeks to seconds, happening as their cars drive around. In addition, they can augment road maintenance work orders in IBM Maximo with better insights and immediate action enabled by BrainMatter.
We also recently won the Schiphol Pointcloud Innovation Challenge, which challenged participants to harness data from 3D/Pointcloud scanners to deliver new or improved products and services to people working at Schiphol Airport.
In just seven days, we developed a solution that monitors maintenance-related assets in Schiphol Terminal 1 nonstop. Members of our team taught BrainMatter to recognize approximately 25,000 assets in nearly 2,000 high-resolution panoramic images with an effort of only a few hours each day.
The judges scored us the maximum number of points for showcasing immediate value for Schiphol and its main contractors, high feasibility and applicability, scalable and intuitive UI, along with a strong commercial approach. The solution will replace manual inspections with automated alerts through tickets when assets such as lighting, smoke detectors and surveillance cameras require maintenance. Consequently, Schiphol and its main contractors can improve their first-time fix rate, enhance their ability to exchange information, reduce costs and better identify risks.
These are just two instances where we’ve delivered tangible value with help from IBM Power Systems. From manufacturing to real estate, airports to logistics, at BrainCreators we know that even when offloading tasks to machines–it is ultimately all about people. It is about curiosity and linking that to experience. About opening the windows to the world of AI and machine learning. All of us live and learn . . . so let’s harness the power of our collective brains.
Learn more about the BrainMatter platform here.
(This article was originally published on the IBM IT Infrastructure blog.)